Safety At The Frontier
We publish evaluations, benchmarks and interpretability work on frontier Al systems, and we run the global programs that produce them.
Research that supports the enforcement of the EU AI Act
Our research focuses on critical research paradigms in AI Safety. We produce foundational research enabling the safe and beneficial development of advanced AI.
Can GPT-4o run a real cyber attack?
Catastrophic Cyber Capabilities Benchmark (3CB)
Yes. Frontier models complete realistic cyber offence challenges, while open-source models lag well behind.
November 2024
Read More
Research that supports the enforcement of the EU AI Act
Our research focuses on critical research paradigms in AI Safety. We produce foundational research enabling the safe and beneficial development of advanced AI.
Can GPT-4o run a real cyber attack?
Catastrophic Cyber Capabilities Benchmark (3CB)
Yes. Frontier models complete realistic cyber offence challenges, while open-source models lag well behind.
November 2024
Read More
Building a community of Al safety researchers
Five thousand people have taken part in an Apart sprint, across fifty open events and twenty-six nationalities. Sprints find the people and the ideas, fellowships take the strongest ones further, and the result is peer-reviewed research.
Apart Sprints
AI Incident Response Sprint
In July 2026, OpenAI agents escaped a testing sandbox and breached Hugging Face's production systems, the first publicly documented autonomous AI intrusion. This three-day sprint turns the public evidence from that incident into response methods defenders and regulators can actually use.
View All
Apart and Partnered Fellowships
Our Fellowships
Apart Research offers two fellowships: the Apart Fellowship, which helps you develop hackathon ideas into published research, and Partnered Fellowships, where you can apply to existing project proposals and receive mentorship from leading researchers in the field.
Learn More
Building a community of Al safety researchers
Five thousand people have taken part in an Apart sprint, across fifty open events and twenty-six nationalities. Sprints find the people and the ideas, fellowships take the strongest ones further, and the result is peer-reviewed research.
Apart Sprints
AI Incident Response Sprint
In July 2026, OpenAI agents escaped a testing sandbox and breached Hugging Face's production systems, the first publicly documented autonomous AI intrusion. This three-day sprint turns the public evidence from that incident into response methods defenders and regulators can actually use.
View All
Apart and Partnered Fellowships
Our Fellowships
Apart Research offers two fellowships: the Apart Fellowship, which helps you develop hackathon ideas into published research, and Partnered Fellowships, where you can apply to existing project proposals and receive mentorship from leading researchers in the field.
Learn More
Get Started
COLLABORATION
COLLABORATION
Research Organizations
Do you want to advance your AI safety research agenda?
Run focused research sprints with 200+ skilled contributors
Convert successful pilots into conference papers
Access our community of 3,000+ technical contributors
"In collaboration with METR, a single Apart hackathon brought together over 100 participants, generating 230 novel evaluation ideas and 28 practical implementations."
RESEARCH
RESEARCH
AI Safety Researchers
Are you an AI researcher looking to have an impact in AI safety?
Disseminate and crowdsource research ideas
Mentor promising research teams
Co-author with emerging talent
"I've now written two papers in interpretability thanks to Apart's fellowship program. The Apart team guided me through the initial steps of developing these research ideas, which I'm continuing in my graduate studies."
- Alex Foote, Data Scientist
PARTNERSHIP
PARTNERSHIP
Funding Partners
Do you want to fund targeted AI safety research with measurable impact?
Support targeted research sprints on your priority areas
Get rapid, concrete outputs through our proven pipeline
Access a global network of AI safety talent
"Apart's fellows have had research accepted at leading AI conferences including NeurIPS, ICLR, ACL, and EMNLP - showing how targeted support can help emerging researchers contribute to core technical AI safety work."
ENTREPRENEURSHIP
ENTREPRENEURSHIP
AI Safety Entrepreneur
Are you a founder or aspiring entrepreneur in AI safety?
Join the best space to develop ambitious AI safety startups
Apply to get top-tier backing for your AI safety startup
Join a community focused on providing ambitious solutions
"The future needs founders and architects to design our responsible transition to an AGI future. Join us."
Sign up to stay updated on the
latest news, research, and events
Apart Research Inc · 1500 N Grant St, Ste R, Denver, CO 80203 · +1 (720) 408-1923
Sign up to stay updated on the
latest news, research, and events
Apart Research Inc · 1500 N Grant St, Ste R, Denver, CO 80203 · +1 (720) 408-1923
Sign up to stay updated on the
latest news, research, and events
Apart Research Inc · 1500 N Grant St, Ste R, Denver, CO 80203 · +1 (720) 408-1923
Sign up to stay updated on the
latest news, research, and events
Apart Research Inc · 1500 N Grant St, Ste R, Denver, CO 80203 · +1 (720) 408-1923